Dr. phil. Gisela HagbergSenior Research Scientist
Clinical studies at Ultrahigh Magnetic Field strengths (UHF):
a window into Brain Microstructure
The advent of ultrahigh magnetic field strengths has boosted the field of neuroanatomy through the remarkable high level of detail that can be achieved in vivo. Fine-grained variations in MR contrast is linked with the underlying neurobiology and can be used to map brain tissue microstructure related to neuropathology as well as cyto- and myeloarchitecture. The advent of UHF thus opens up new possibilities to characterize healthy tissue, and investigate disease-related alterations of brain microstructure, eventually bringing in vivo histology with clinical relevance within reach.
Yet, with increasing field strength, methodological challenges in terms of increased inhomogeneity of the static field and the transmit field arise. Therefore, in order to achieve a high reproducibility in vivo, such factors need to be addressed
You'll find out more about on-going clinical projects .
A core of quantitative MRI tools
for clinical and neuroscientific in vivo research at 9.4T has been established. Besides mapping of the transmit field, necessary to correct for the strong inhomogeneity, they consist of MP2RAGE-based mapping of the longitudinal relaxation (T1; R1) with isotropic voxel-sizes of 800µm down to 300µm; a multi echo 3D GRE sequence; with voxel-sizes of 375x375x800µm down to an isotropic voxel size of 400µm mainly for mapping the effective transverse relaxation (T2*; R2*), but that can also be used for quantitative susceptibility mapping (QSM); and a high-resolution acquisition weighted, single-echo 3D GRE sequence; with voxel-sizes down to 113x118x600µm used for QSM.
The whole protocol has a total duration of 45min and is therefore suited for clinical examinations of patients using a multi-modal imaging approach but without too harsh constraints on patient compliance. For validation purposes, this core of sequences has been complemented with single slice inversion-recovery mapping of T1 using EPI, and CPMG based mapping of T2, with an echo train consisting of 32 echoes and a minimum inter-echo time of 9ms, without exceeding SAR limits.
These protocols have been used to elucidate amongst other age-depndent variation in healthy subjects, the anatomical detail of the superior colliculus, beta-amyloid plaques in Alzheimer's and in post-mortem tissue.
The severe inhomogeneity of the transmit field poses a limit for quantitative studies at 9.4T. We found that this factor not only affects the MR signal excitation, but also impacts inversion efficiency, despite the use of adiabatic pulses. Such pulses are generally assumed to be immune to variations in B1, but through Bloch simulations and experiments we could confirm studies from the early 90 demonstrating the vulnerability of that assumption (Hagberg et al., 2017). These findings further led us to propose an efficient correction for T1 (R1) mapping at 9.4T that tackles this issue and yields a high in vivo test-retest variability (< 1%) and low coefficient-of-variation (CoV<3% across 25 subjects at the level of extended regions-of-interest). Consistent with previous studies, we found that the primary sensory and motor areas had the greatest quantitative R1 values, in line with their greater myelin content with respect to other brain areas, and observed that the quantitative MRI parameters measured at 9.4T have a greater dynamic range than at 3T (Fig. 1). Nevertheless we could observe that the voxel wise CoV reached levels beyond 15% in grey matter regions, being several fold higher than in white matter voxels. This is interesting in view of the difficulty to detect myelin-related signals outside the white matter at 3T (Groeschel et al., 2016). Upon closer scrutiny we found that this effect derives from age-related variation in R1. The observed dynamics of MR-parameters across the human life-span could be explained in a model that takes into account decreasing grey matter volume fractions, increased iron levels and increased myelination of the cortical fibers with age (Fig. 2 and Hagberg et al., 2017b).
With further advancing age, several changes of the brain microstructure occur that may impact quantitative MRI measures. With age, the likelihood of presenting ?-amyloid deposits increases, with prevalence peaking at age 70, even in a cognitively healthy population (Jack et al., 2017). In patient populations showing mild cognitive impairment, early identification of the presence of such deposits is essential since patients without them have a different clinical course towards non-Alzheimer dementia. Recently we found that QSM of post mortem samples can detect single ?-amyloid plaques in Alzheimers at 14.1T, but only when voxel-sizes (37-50µm) on the order of the size of single plaques were employed (Fig. 3 and Tuzzi et al., 2016; 2017). This finding poses important questions regarding the best imaging protocols to use for in vivo MRI and the resolution necessary to answer clinical questions at UHF, since pathologyrelated effects of ?-amyloid can be detected at high field but with coarser sampling (van Rooden et al., 2016).
A particularly challenging area for MRI research are the function and the anatomy of the deep brain nuclei located in the midbrain, characterized by a limited available signal-to-noise-ratio and a strong variation in background magnetic susceptibility. The superior colliculus (SC) is one of these brain nuclei and is a layered structure which serves as a gateway for several sensory modalities. Up-to-date, the layering pattern could only be identified in histology, or in vivo by functional challenge. We recently showed that layer specific fMRI activations of the superior colliculus (SC) can be detected at 9.4T (Loureiro 2017a). More recently (Loureiro 2016; 2017b) we found that the combined use of several quantitative MRI parameters provides a mean to study layer specific anatomy of the SC in vivo with features that are sensitive to the local microstructure that are reproducible across subjects. Moreover we observed that the combined use of several MRI parameters enabled identification of several anatomical structures within the midbrain areas (Fig. 4). Along with these measurements we employed MRI of post mortem samples at 14.1T and polarized light microscopy imaging to validate our observations regarding detectability of white matter fascicles in the brain stem. We are currently pursuing research activities in this direction with the aim to pin-point the source behind the obtained quantitative MRI measures and have extended our focus to include signal variations within and beyond the cortical rim in the occipital cortex (Fig. 5).
1985-1988 Institute of technology, Lund (LTH), Sweden
1988-1990 Institute of technology, Lausanne (EPFL), Switzerland. Engineering project: 2D Nuclear Overhauser Effects (G. Bodenhausen)
1990 MSc in Engineering physics, EPFL and LTH. Thesis: Measurement of esophagal tumor depth by Laser-Induced Fluorescence (A.Châtelain)
1990-1993 PhD trainee, University of Basel, Switzerland
1993 PhD in Biophysics, University of Basel, Switzerland. Thesis: Metabolism of the human brain investigated by 1H-MRS (J. Seelig)
2007 Medical Physics Specialization, University La Sapienza, Rome, Italy. Thesis: Detection of epileptogenic foci by neuronal current MRI (B. Maraviglia)
POSITIONS HELD at MPI
2010-2012 Post-doctoral Fellow, Logothetis Group, MPI for Biological Cybernetics, Tübingen, Germany
2012-current: Research Scientist/Project leader, Scheffler Group, MPI for Biological Cybernetics and University hospital Tübingen, Germany
68articles in peer-reviewed journals, co-author of 5 book-chapters, h-index:26 ; (see: ); i10: 52 (see Google Scholar)
University of Tübingen, Germany; ECM continuing education in medicine: BHC - Rome, Fondazione Santa Lucia Rome; Lecturer Universita Tor Vergata, Co-supervisor of five Msc, one Medical Physics Specialization thesis and four PhD theses.
Italian Ministry of Health: RF00.97.; RF00.99; RF05.103; University Tübingen CIN/Pool 2012-11; EU-LAC 2017
US Patent 7430313: Methods using recurrence quantification analysis to analyze and generate images. Zbilut JP, Sirabella P, Bianciardi M, Hagberg G, Colosimo A,Giuliani A.
Brevetto RM 00 A 000160, 31.3.2000: Method and equipment for generation of quantitative T2* maps for production of Magnetic resonance images in real-time GE Hagberg, S Posse, JN Sanes.
Karolinska Institute - Stockholm, CNR - Genova, International School on Magnetic Resonance and Brain Function Erice, Strasbourg, France, Radboud University, Nijmegen, The Netherlands; Workshops and congresses of AIFM; ISMRM; EFOMP
American journal of Neuroradiology, Journal of Neurochemistry, Magnetic Resonance in Medicine, Medical Physics, Nature Medicine, NMR in Biomedicine, Neuroimage, Neuroradiology, Plos oNE, Abstract review: ISMRM 2006, 2009, 2012-2014; 2016 Brain PET, 2007, 2009, 2011, 2014
Memberships, organization of Congresses and Workshops
Italian Association for Medical Physics (AIFM, since 2005). Board member organizing and scientific committees: Italian Federation of Radiation Research (FIRR) 2009, 2011; AIFM: 2011, 2013; ISMRM/AIFM/EFOMP: 2010